Including Knowledge Base Research Articles

AJE: Attention Mechanism for Entity-relation Joint Extraction

Joint extraction of entities and relations is an significant issue of information extraction, which is very helpful for many downstream tasks, including knowledge base construction, question answering, and biomedical text diagnosis[1], etc. The common approach of existing models is to extract the subject and the relation first, then compute the subject and the relation to obtain the object, and finally, the triplet is judged. However, such an approach cannot efficiently handle of information extraction, and the results are not very good for Subject-Object Overlap (SOO) case. In this paper, a joint entity-relation extraction method AJE is proposed based on dot-product attention mechanism. The method first maps subject, object and relationship into three matrices of Q, K and V. After that, the attention weighting is achieved on these three matrices and the output vector is used to determine whether the triple is correct or not. The F1-score is used in experiments to show that the proposed model is more efficient than the current existing ones. It also has better results in handling other cases such as SOO, multi-triple problem, etc.

Power Dispatching Voice Backtracking System Based on Knowledge Base

With the rapid development of the scale and structure of the power grid, the power dispatching center has built a complete dispatching exchange communication network, which can meet the daily dispatching communication needs, but various problems are also emerging in the dispatching process. Aiming at the problem of high integration of physical power grid and information system, this paper proposes a voice backtracking system for power dispatching based on knowledge base, which includes knowledge base establishment, knowledge storage, knowledge question answering sorting and knowledge base front-end question answering. Firstly, the basic model of power dispatching knowledge processing is trained, and then the knowledge is extracted from the domain corpus, and finally, the knowledge question answering is realized by using the ranking model. Through the experimental comparison, it is proved that the accuracy of the method in this paper is higher than that of the method in the postprocessing by adding the MRC model in the pre-processing of knowledge extraction, and it is better than other methods in the comparative test. It provides technical support for power system voice dispatching management and provides a reference for other areas of voice management.

Evaluating and Enhancing the Robustness of Sustainable Neural Relationship Classifiers Using Query-Efficient Black-Box Adversarial Attacks

Neural relation extraction (NRE) models are the backbone of various machine learning tasks, including knowledge base enrichment, information extraction, and document summarization. Despite the vast popularity of these models, their vulnerabilities remain unknown; this is of high concern given their growing use in security-sensitive applications such as question answering and machine translation in the aspects of sustainability. In this study, we demonstrate that NRE models are inherently vulnerable to adversarially crafted text that contains imperceptible modifications of the original but can mislead the target NRE model. Specifically, we propose a novel sustainable term frequency-inverse document frequency (TFIDF) based black-box adversarial attack to evaluate the robustness of state-of-the-art CNN, CGN, LSTM, and BERT-based models on two benchmark RE datasets. Compared with white-box adversarial attacks, black-box attacks impose further constraints on the query budget; thus, efficient black-box attacks remain an open problem. By applying TFIDF to the correctly classified sentences of each class label in the test set, the proposed query-efficient method achieves a reduction of up to 70% in the number of queries to the target model for identifying important text items. Based on these items, we design both character- and word-level perturbations to generate adversarial examples. The proposed attack successfully reduces the accuracy of six representative models from an average F1 score of 80% to below 20%. The generated adversarial examples were evaluated by humans and are considered semantically similar. Moreover, we discuss defense strategies that mitigate such attacks, and the potential countermeasures that could be deployed in order to improve sustainability of the proposed scheme.

Air Pollution - Sources and Classification

Air pollution is a growing and alarming issue globally and needs to be addressed at all levels. Its understanding in terms of sources and classification is of utmost importance. The present paper deals with types of air pollution along with sources, classification, and definition of air pollution. It covers general aspects including knowledge base for the students and others to understand the emerging issues related to air pollution.

ONTOLOGY-BASED DEVELOPMENT OF DIAGNOSTIC INTELLIGENT SYSTEMS

Development of problem-oriented intelligent systems for solving semistructured problems remains a topical problem. The solution efficiency of this problem can be increased by improving approaches (methodologies) or creating specialized software. The paper describes an approach for the ontology-based development of diagnostic intelligent systems (including knowledge bases). A model-driven approach is used as a methodological basis with a subsequent redefinition of the main stages, models, transformation rules and means of implementation. The proposed approach is used for development of diagnostic systems in petrochemistry. Diagnosis is considered as the definition of the object’s state, including: monitoring the state, finding a location and determining the causes of a failure, forecasting the state. The objects to be diagnosed are presented as a sequence (chain) of states, where the state is a set of parameters and their values at a given time. The features of the proposed approach implementation are the following: the use of ontology as a computation independent model that describes the features of the problem; the use of the Rule Visual Modeling Language to present the platform independent and platform dependent models; the use of C Language Integrated Production System as a programming language and a target platform. An example of the approach application by means of software prototypes called Knowledge Base Development System and Personal Knowledge Base Designer is described.

MATHEMATICAL SUPPORT OF THE INTELLIGENT INFORMATION SYSTEM OF ASSESSING THE OBJECT STATE

At present, information technologies (IT) are intensively used all over the world in various sectors, and today medical institutions cannot do without them when organizing the process of medical diagnostic. The IT efficiency is determined by the degree of their intellectualization that is by including knowledge bases as their component and by the transition from data processing to the processing of knowledge. The efficiency of making decisions in various areas of activity is determined by the quality and quick delivery of information. Medicine constitutes no exception in this sense. The advanced level of computer technology, applied tools, diagnostics on the basis of automated systems of decision support made it possible to solve the tasks of assessing the state of the object at a qualitatively new level. The subject matter of this study is to ensure the mathematical support of the intelligent information system (IS) of assessing the state of the object. The object is understood as a patient who came through a myocardial infarction (MI). The goal of the study is to develop mathematical support of the intelligent IS of assessing and predicting a patient’s condition. To achieve the stated goal, the following tasks were solved: statistically valid and uncorrelated signs were specified; these signs enable distinguishing the group of patients who survived from those who died, “decisive rules” were formulated for predicting the MI clinical outcome. In the process of the study, the mathematical IT of assessing the state of the object was developed. The following result was obtained: the suggested mathematical models for predicting the outcome of myocardial infarction that were developed with the use of the method of discriminant function and took into account human blood values can prevent sudden coronary death and improve the diagnostic efficiency. Conclusions. Mathematical models were developed to predict the state of the object in the event of uncertainty. The use of the developed mathematical models enables improving the accuracy of predicting the state of the object in a real-time environment and in the early stages of the disease development by 4.2% and 10%, and applying adequate preventive and therapeutic and rehabilitation measures in time as well as preventing sudden coronary death. The developed mathematical models were tested.

Knowledge verification for long-tail verticals

Collecting structured knowledge for real-world entities has become a critical task for many applications. A big gap between the knowledge in existing knowledge repositories and the knowledge in the real world is the knowledge on tail verticals (i.e., less popular domains). Such knowledge, though not necessarily globally popular, can be personal hobbies to many people and thus collectively impactful. This paper studies the problem of knowledge verification for tail verticals ; that is, deciding the correctness of a given triple. Through comprehensive experimental study we answer the following questions. 1) Can we find evidence for tail knowledge from an extensive set of sources, including knowledge bases, the web, and query logs? 2) Can we judge correctness of the triples based on the collected evidence? 3) How can we further improve knowledge verification on tail verticals? Our empirical study suggests a new knowledge-verification framework, which we call F acty , that applies various kinds of evidence collection techniques followed by knowledge fusion. F acty can verify 50% of the (correct) tail knowledge with a precision of 84%, and it significantly outperforms state-of-the-art methods. Detailed error analysis on the obtained results suggests future research directions.

Study on Fault Diagnosis Expert System of Aircraft Power-Supply System

We always regard aircraft power supply system as the "blood system" of aircraft. It plays a very important role in aircraft work. In view of its fault diagnosis present situation and in order to improve fault diagnosis efficiency, we put forward to use expert system development tool CLIPS to build up fault diagnosis expert system. In this paper, we choose the power-supply system of Cirrus SR20 as diagnosis object, and choose CLIPS as development tool to build up knowledge base and inference engine. By using Eclips development platform to write interface programs and using CLIPS JNI to call CLIPS programs we successfully complete the expert system total performance including knowledge base, inference engine and interface.

Software test bench for mathematical models debugging and verification

The problem of software test bench development for debugging and verifications of the mathematical models used in various observations and control tasks is considered. The test bench includes knowledge base of mathematical models, the program generator of test signals and the quality of models functioning analyzer. Realization of the test bench is carried out on the basis of javatechnologies.

Design and realization of a modular architecture for textual entailment

AbstractA key challenge at the core of many Natural Language Processing (NLP) tasks is the ability to determine which conclusions can be inferred from a given natural language text. This problem, called theRecognition of Textual Entailment (RTE), has initiated the development of a range of algorithms, methods, and technologies. Unfortunately, research on Textual Entailment (TE), like semantics research more generally, is fragmented into studies focussing on various aspects of semantics such as world knowledge, lexical and syntactic relations, or more specialized kinds of inference. This fragmentation has problematic practical consequences. Notably, interoperability among the existing RTE systems is poor, and reuse of resources and algorithms is mostly infeasible. This also makes systematic evaluations very difficult to carry out. Finally, textual entailment presents a wide array of approaches to potential end users with little guidance on which to pick. Our contribution to this situation is the novel EXCITEMENT architecture, which was developed to enable and encourage the consolidation of methods and resources in the textual entailment area. It decomposes RTE into components with strongly typed interfaces. We specify (a) a modular linguistic analysis pipeline and (b) a decomposition of the ‘core’ RTE methods into top-level algorithms and subcomponents. We identify four major subcomponent types, including knowledge bases and alignment methods. The architecture was developed with a focus on generality, supporting all major approaches to RTE and encouraging language independence. We illustrate the feasibility of the architecture by constructing mappings of major existing systems onto the architecture. The practical implementation of this architecture forms the EXCITEMENT open platform. It is a suite of textual entailment algorithms and components which contains the three systems named above, including linguistic-analysis pipelines for three languages (English, German, and Italian), and comprises a number of linguistic resources. By addressing the problems outlined above, the platform provides a comprehensive and flexible basis for research and experimentation in textual entailment and is available as open source software under the GNU General Public License.

Data Process of Diagnose Expert System based on Neural Network

Engine fault has a high rate in the car. Considering about the distinguishing feature of the engine, Engine Diagnosis Expert System was investigated based on Diagnosis Tree module, Fuzzy Neural Network module, and commix reasoning module. It was researched including Knowledge base and Reasoning machine, and so on. In Diagnosis Tree module, the origin problem was searched in right method. In which module distinguishing rate and low error and least cost was the aim. By means of synthesize judge and fuzzy relation reasoning to get fault origin from symptom, fuzzy synthesize reasoning diagnosis module was researched. Expert knowledge included failure symptom, engine system failure and engine part failure. In the system, Self-diagnosis method and general instruments method worked together, complex failure diagnosis became efficient. The system was intelligent, which was combined by fuzzy logic reasoning and the traditional neural network system. And it became more convenience for failure origin searching, because of utilizing the three methods. The system fuzzy neural networks were combined with fuzzy reasoning and traditional neural networks. Fuzzy neural network failure diagnosis module of system, as a important model was applied to engine diagnosis, with more advantages such as higher efficiency of searching and higher self-learning ability, which was compared with the traditional BP network

An Expert System for Aided Design of Rice Transplanter Chassis Systems

The paper presents a structure and functions of an expert system for aided design of rice harvester chassis systems. It was developed on basis of a detailed analysis of the design process, including knowledge bases regarding chassis systems automated design, a community of parameterized parts and assemblies and interference engine. In the creation of the system, Visual C++, Pro/Engineering and Oracle database were used. This system is characterized by a rule-oriented representation of knowledge, forward chaining inference methods, and graphic representation of the design results by parameterized technology. The industrial application of this system proved its high reliability and accuracy.

A decision support system for lower back pain diagnosis: Uncertainty management and clinical evaluations

Lower back pain (LBP) is a common medical problem that deprives many individuals of their normal lifestyles and keeps them from routine activities. Diagnosing LBP is challenging because it requires highly specialized knowledge involving a complex anatomical and physiological structure as well as diverse clinical considerations. Although a handful of studies have proposed or developed systems to support LBP diagnosis and improve knowledge sharing, these systems have limited scope, lack systematic evaluations, and/or ignore diagnoses that consist of multiple parts (i.e., decision outcomes), each of which corresponds to a particular medical condition, disease, or abnormality. In this study, we design, implement, and evaluate a Web-based decision support system that employs an intuitive and easy-to-use framework to assess the patient's information and recommend a diagnosis consisting of one or multiple parts. Our system design addresses the challenging characteristics of a LBP diagnosis and uses verbal probability estimation to represent and reason about the associated uncertainty. Our evaluations are systematic, including knowledge base verification, system validation using a modified Turing test, and clinical efficacy assessment involving 5 clinicians and 180 real-world cases collected from geographically dispersed clinics. Our evaluation design is more thorough than those used by most previous studies, and the proposed system is relatively ready for clinical deployment. Therefore, this study both contributes to decision support systems research and has advanced clinical support for LBP diagnosis. In light of some of the limitations of this study, we also identify and discuss several areas that need continued investigation.

Investigation of an Expert System for the Condition Assessment of Transformer Insulation Based on Dielectric Response Measurements

The need for economic, reliable, and effective delivery of electric power has lead to the search for fast, efficient, and effective methods for diagnosing the insulation of high-voltage (HV) equipment in the power industries. The recent dielectric techniques that have been carefully considered by major industries for transformer insulation condition assessment are the recovery voltage method (RVM) and the polarization and depolarization current (PDC) measurement. However, due to the complexity of the transformer insulation structure and various degradation mechanisms under multiple stresses, insulation condition assessment is usually performed by experts with special knowledge and experience. In this paper, an expert system (ES) is developed, which imitates the performance of a human expert, to make the complicated insulation condition assessment procedure accessible to plant maintenance engineers. The structure of the ES is described in detail including knowledge base, inference engine, and human-computer interface. Examples of the application of the ES are also presented to confirm that the system can provide accurate insulation diagnosis.

Knowledge acquisition for expert system development

The recent proliferation of expert system development projects has focused the attention of the AI community on the problem of knowledge acquisition. Traditional programming skills possessed by the majority of knowledge engineers are found to be of little use when faced with the problem of extracting potential knowledge base information from a human expert. This paper deals with those topics and situations which have so often been described as the ‘bottleneck’ of expert system development. The concepts of expert and expertise are explored; the paradox of expertise is examined. Those interviewing methods found to be most effective for knowledge acquisition are discussed. A broad gamut of knowledge acquisition considerations are covered in detail including knowledge base structure, conversion of knowledge into rules, and refinement of the knowledge base.

A method on software development and maintenance support with reference to documentation (1st report, Basic concept of the suport tool and contradiction detection methods in logic)

A basic concept of a software support tool for the production, maintenance and updating is presented employing a structure of the expert system interacting with both software and documentation. This support tool can include knowledge bases, data bases of target softwares, and also other related knowledge bases. As a first step toward this direction, a small sized support tool for software development has been developed by introducing software grammers, syntactics and semantics in a small world as rules and documents analizer. Three types of the contradiction detection method are incorporated in this system so as to detect contradictions in documentations and an additionally inserted knowledge in the expert system itself, based on the semantics and the knowleges, which are applicable for many expert systems. This system has two types of checking methods, such as a software based checking and a documentation based checking, depending upon the applications. Some experiments of the system have been carried out as a man-machine interface.